When third grade teacher Brad Upshaw brought an Amazon Echo into his classroom, students spotted it immediately. Like reporters at a press conference, they swarmed the device and began shouting out questions.
“It was plugged in behind my interactive whiteboard,” says the teacher at Los Angeles Unified School District’s Vanalden Avenue Elementary School “It was barely visible under the screen, and students did not know that I had brought it.”
Their swiftness in recognizing the artificial intelligence (AI) device – and their fearlessness in interacting with it – were eye-opening. Kids, he believes, are ready for AI.
“As it is, whenever the students or myself get stuck needing something answered in class during the day, the students simply suggest, ‘Just ask Siri,’” he says. “I have no doubt that this year’s students, and all that follow, will have no barriers to interacting with AI.”
In a way, the AI revolution crept up when most of us weren’t looking. For decades, researchers have been extolling the potential feats of artificial intelligence, but the technology wasn’t there yet. Now it suddenly is. We can talk to our devices and get our questions answered. We have smart homes, smart cars, smart appliances and smart speakers. Algorithms predict what we’ll like, anticipate our behavior and even diagnose our medical conditions – all with frightening accuracy.
Artificial intelligence has already infiltrated our lives. It just doesn’t look like we expected.
From schools, however, it remains strangely absent. Machine learning (the technology that allows AI to grow smarter) has enriched nearly every industry, but almost none of those advances have significantly improved education, points out Barbara Kurshan, executive director of Academic Innovation at the University of Pennsylvania. Today’s students can expect to interact with AI both on the job and in their home lives. Yet aside from some isolated experimentation, momentum for AI in the classroom seems to have largely faded.
But not for long. In the next year or two, AI is going to explode into K-12 schools, predicts Hall Davidson, senior director of Discovery Education and a member of the ISTE Board of Directors.
“Whether you’re doing it now or not, it absolutely is going to be coming,” Davidson says. “We don’t want educators to be taken by surprise.”
Talking to the walls
Devices like the Echo or the Google Home have launched us into a sci-fi reality where the fastest way to get answers is to simply ask the room. As voice recognition technologies like Siri and Alexa get smarter, looking up information the old way – by launching a web browser and typing in a search term – will soon seem clunky, archaic and unreasonably slow.
Davidson sees the potential of voice-recognition AI to support students in reaching higher levels of learning and thinking. By asking intelligent questions and thinking out loud, students can use these devices to gain new insights and propel their explorations.
“We’re entering a world where conversation with machines is going to be both enlightening and more efficient in many areas,” he says. “I think it really should be part of the school environment when they walk in the door. We want people to be able turn to the wall and ask questions, and just have it spit out they answer so they can go on with their thinking.”
Decades ago, early AI researchers described a technology called a lifelong learning companion that would follow students year after year, getting to know their interests, skills and learning habits and responding with questions or feedback to keep them engaged. It’s coming – but will students be ready to leverage AI as empowered learners? Harnessing the power of Siri to solve problems, build knowledge and guide their own learning prepares them for this future while helping them meet the ISTE Standards for Students.
The ability to interact effectively with voice-activated AI may even become a necessary job skill for many students. A doctor, for example, could find substantial value in conversing with a device that has read every article on a particular medical subject.
“A doctor can’t read all that themselves. There’s just no time to do it,” Davidson says. “You can’t replace a doctor with AI, but it would be nice to have somebody sitting there who has read every article in every journal.
“If we want kids to be innovative and really master that world, we might as well start them now.”
Is AI the new TA?
The best way to teach students how to interact with artificial intelligence is, of course, to model it. Experts describe a not-too-distant future in which teachers will work side by side with AI assistants that will augment their human expertise with real-time, data-driven learning recommendations.
One of machine learning’s greatest potential impacts on education is its ability to multiply the intelligence at the teacher’s disposal.
“The classroom typically has one intelligence – one teacher – managing 30 kids,” says Lehigh University professor Scott Garrigan, who presented a session on “What to Expect From Artificial Intelligence in K-12” at ISTE 2017.
“AI is the first technology that can make the kinds of judgments that we used to need teachers to make. With AI, you have multiple intelligences in the room and multiple sources of judgment.”
Adaptive learning programs like Khan Academy allow educators to achieve new levels of personalized learning by using algorithms to assess a student’s knowledge level, identify gaps and adjust the instruction accordingly. Through a combination of voice recognition and eye tracking, AI can identify who is doing or saying what in a group activity and pinpoint which learners are focusing on which learning resources at any given time. It can make inferences about a student’s learning patterns, interests, emotional state and more – and then it can make decisions based on them.
Perhaps even more critically, it can give educators deeper insight into the learning process. Pearson, an education publishing and assessment service, calls artificial intelligence “a powerful tool to open up what is sometimes called the ‘black box of learning,’ giving us deeper, and more fine-grained understandings of how learning actually happens.” For example, AI can track the micro-steps students go through to learn a specific subject, such as physics, to help teachers design more effective lesson plans.
So far, adaptive learning has been used primarily as a one-on-one tutor. Research suggests these types of intelligent tutoring systems (ITS) can achieve “remarkable increases in student learning over traditional classroom instruction in the real world,” helping students achieve greater competency in a shorter amount of time, says Bill Ferster, author of Teaching Machines: Learning from the Intersection of Education and Technology.
In one study, students using an ITS reached the same level of competency in 20-25 hours of instruction as students who spent more than four years in traditional training. Similarly, researchers found that students who used an algebra I tutor performed 85 percent better on assessments of complex problem-solving skills. In another review, intelligent tutoring systems were associated with higher outcome scores across a wide range of learning conditions.
“There’s no question; the research is strong that if designed properly, these intelligent systems can teach kids better and faster than almost any other technique,” Ferster says.
That doesn’t mean AI will replace teachers. Like any other technology, it’s only as good as the person using it, and classrooms will always need human brains to orchestrate meaningful learning experiences. But intelligent software can automate a lot of the tasks only teachers used to be able to do, like grading, assessments and classroom management. It can even make decisions about how to arrange student groups to optimize collaborative learning.
“A teacher has their hands full and cannot meet the needs of every kid in every individual way,” Garrigan says. “AI can make decisions on data so the teacher doesn’t become a bottleneck to the kids’ learning.”
Opening the door for AI
What will AI in the classroom ultimately look like? No one knows yet.
Right now, “it looks like nothing,” Davidson says. “As we are looking at it now, any integration at all would be a success.”
There are obstacles, of course. Parents and teachers alike are often wary of AI in the classroom – it’s only natural, given that the World Economic Forum predicts automation will eliminate at least 5 million jobs worldwide by 2020. Those who do support it face more practical barriers, like basic classroom design.
Since most of the adaptive learning programs available today are designed for one-on-one tutoring, they don’t translate well into the typical classroom environment where one teacher is engaging 30 or more kids at once.
“You’ve got an architecture problem in the relationship between students, the teacher and the classroom,” Ferster says. “It’s a challenge for classrooms to use these new technologies because they’re not set up for the whole class. That’s an impediment to any kind of progress.”
Adaptive learning makes the most sense in flipped or blended learning environments, where students routinely work independently on connected devices. As more classrooms transition to these models, we may start to see these technologies applied on a larger scale.
Until then, Davidson encourages teachers to start small. With all of the iPhones entering classrooms these days, it’s not hard to weave Siri into a lesson. And it only costs $49 to get a classroom Echo, provided you can get it past your school’s filters to connect to the internet. (That’s one problem Upshaw ran into when hooking up his Echo.)
At minimum, teachers can use emerging technologies to teach students how to be savvy consumers of information by experimenting with asking different types of questions exploring the limits of current artificial intelligence.
“What we really want is for kids to understand the power of this tool and find ways to use it that we wouldn’t have thought of,” Davidson says. But, he adds, it’s also important to show them that AI devices “are fools and can be fooled.” Although they’ll get smarter as more people use them, AI apps like Siri or Alexa are still easy to stump. The ability to ask intelligent questions will make a difference in the quality of answers students get back.
“A good teacher asks very good questions,” Upshaw says. “The important task for K-12 teachers is to guide students to forming the essential questions for their learning. I’m looking forward to this new partnership between the curious young mind and the interactive databases available.”
Nicole Krueger is a freelance writer and former newspaper reporter. She writes about education technology and the transformation of learning.